cyclo_lab
A simulation research repository providing reinforcement- and imitation-learning environments for ROBOTIS robots, built on NVIDIA Isaac Lab, with sim-to-real deployment support.
cyclo_lab provides simulation environments, configuration tools and task definitions for training reinforcement learning and imitation learning policies on ROBOTIS robots, using NVIDIA Isaac Sim's GPU-accelerated physics and Isaac Lab's modular RL pipeline. It depends on Isaac Lab 2.2 or newer and ships a Docker setup that bundles Isaac Sim, Isaac Lab, CycloneDDS and a separate LeRobot environment.
The repository includes example reach and lift tasks for the OMY robot arm that can be trained and then played back in simulation, and it supports sim-to-real deployment so a policy trained in Isaac Lab can run on the physical robot.
It is written in Python and licensed under Apache-2.0.